Dagger is a modern, open-source platform for automating software delivery pipelines. Created by Solomon Hykes (the creator of Docker), Dagger replaces fragile shell scripts and vendor-specific CI/CD YAML with real, type-safe code that runs identically on your laptop, in your CI server, or directly in the cloud.

How It Works

Dagger provides a complete execution engine built on containers. You write pipeline logic using native SDKs available in Go, Python, TypeScript, PHP, Java, .NET, Elixir, and Rust. Each SDK is generated from the Dagger API schema, giving you full type safety, IDE autocompletion, and compile-time checks. Pipelines are composed of typed functions that orchestrate containers, filesystems, secrets, git repositories, and network tunnels.

Key Features

Local-First Execution: Run the exact same pipeline locally and in CI. The only dependency is a container runtime like Docker. No more “works on my machine” problems with CI/CD.

Incremental by Default: Every operation is keyed by its inputs using content-addressed caching. Change one file and only affected operations re-run. This works automatically across local runs and CI.

Reusable Modules: The Dagger ecosystem includes a rich registry of community modules (called Daggerverse). Import modules to handle common tasks like building Docker images, running linters, deploying to Kubernetes, or publishing to package registries.

Built-in Observability: Every operation emits OpenTelemetry spans. The CLI includes a live terminal UI for watching execution. Traces can also be exported to Jaeger, Honeycomb, or any OTel-compatible backend.

Interactive REPL: The Dagger Shell provides an interactive environment for composing and debugging pipelines step by step, making it easy to explore APIs and test functions before committing them to code.

Integration Points

Dagger runs in any CI system (GitHub Actions, GitLab CI, CircleCI, Jenkins, etc.) since it only needs Docker. It can also be invoked from Make, Just, or any task runner. The typed artifact system lets you pass objects between SDK languages and modules seamlessly.

Agent Integration

AI coding agents can use Dagger to automate build, test, and deployment workflows programmatically. The type-safe SDKs and interactive shell make it straightforward to generate, modify, and execute pipeline code. Dagger’s container-based isolation ensures agent-generated pipelines run safely and reproducibly.